ElevenAgents: Test-specific tool mocks: Agent simulations and unit tests can now define…
By Steven Van ·
Test-specific tool mocks: Agent simulations and unit tests can now define tool_mock_overrides, keyed by tool ID, to replace shared mocks for one test.
ElevenLabs's ElevenAgents tools now let agent simulations and unit tests define tool_mock_overrides, a map keyed by tool ID that swaps in a mock for a single test instead of the shared one. Mock responses also gain an optional is_error flag (default false), so a test can exercise a tool-failure path.
The same August 3 changelog adds several other changes to ElevenAgents:
- Literal Say node messages can carry text_translations, a map of language keys to translated strings.
- Listing conversations and searching conversation messages both accept a version_id parameter to filter results by agent version.
- A new knowledge_base system tool lets an agent's configuration set enabled retrieval strategies, and RAG queries can pass use_agent_defaults as false to use neutral retrieval settings while keeping the agent's embedding model.
- Agent configuration adds voice activity detection (VAD), currently supporting a background_voice_detection option.
- Custom LLM integrations can now use a websocket API type.
- Widget file-input configuration adds max_files_in_memory (1-30, default 10); files beyond the limit are summarized and released from memory.
- DTMF input configuration adds redact_input, which replaces caller keypad digits in transcripts, conversation logs and analysis, while digits repeated by the agent or passed to tools stay unchanged.
ElevenLabs
AI audio platform for text to speech, voice cloning, and video dubbing across dozens of languages while preserving the original tone and emotion.
View ElevenLabs →